Geoscientific Model Development (Sep 2020)
An exploratory performance assessment of the CHIMERE model (version 2017r4) for the northwestern Iberian Peninsula and the summer season
Abstract
Here, the capability of the chemical weather forecasting model CHIMERE (version 2017r4) to reproduce surface ozone, particulate matter and nitrogen dioxide concentrations in complex terrain is investigated for the period from 21 June to 21 August 2018. The study area is the northwestern Iberian Peninsula, where both coastal and mountain climates can be found in direct vicinity and a large fraction of the land area is covered by forests. Driven by lateral boundary conditions from the European Centre for Medium-Range Weather Forecasts (ECMWF) Composition Integrated Forecast System, anthropogenic emissions from two commonly used top-down inventories and meteorological data from the Weather Research and Forecasting Model, CHIMERE's performance with respect to observations is tested with a range of sensitivity experiments. We assess the effects of (1) an increase in horizontal resolution, (2) an increase in vertical resolution, (3) the use of distinct model chemistries, and (4) the use of distinct anthropogenic emissions inventories, downscaling techniques and land use databases. In comparison with the older HTAP emission inventory downscaled with basic options, the updated and sophistically downscaled EMEP inventory only leads to partial model improvements, and so does the computationally costly horizontal resolution increase. Model performance changes caused by the choice of distinct chemical mechanisms are not systematic either and rather depend on the considered anthropogenic emission configuration and pollutant. Although the results are thus heterogeneous in general terms, the model's response to a vertical resolution increase confined to the lower to middle troposphere is homogeneous in the sense of improving virtually all verification aspects. For our study region and the two aforementioned top-down emission inventories, we conclude that it is not necessary to run CHIMERE on a horizontal mesh much finer than the native grid of these inventories. A relatively coarse horizontal mesh combined with 20 model layers between 999 and 500 hPa is sufficient to yield balanced results. The chemical mechanism should be chosen as a function of the intended application.